Governments Implement AI Policies to Prevent Vendor Lock-in
Governments in the United States, France, and other nations are adopting policies to ensure AI model portability and digital sovereignty to avoid vendor lock-in.
Governments worldwide are implementing policies to prevent vendor lock-in as they integrate artificial intelligence into public infrastructure. In the United States, the Office of Management and Budget issued a 2025 memorandum requiring federal agencies to ensure data and model portability in all AI contracts. This effort aligns with a 2026 review by the Government Accountability Office, which identified vendor lock-in as a live risk to government operations.
France is taking more aggressive steps toward digital sovereignty by migrating 2.5 million civil servants from Microsoft Teams, Zoom, and Webex to a sovereign alternative by 2027. This transition follows admissions that U.S. authorities could access French government data under the Cloud Act, a concern echoed by Australia's banking regulator.
These policy shifts were accelerated by a June incident in which a United States Department of Commerce export-control order forced Anthropic to restrict foreign nationals' access to its Fable 5 and Mythos 5 models. In response, leaders from France, Canada, and the United Kingdom called for diversification of AI providers to maintain national sovereignty.
